library(kableExtra)
library(tidyverse)

font <- "Roboto Condensed"
theme_set(theme_minimal(base_size = 14, base_family = font))
my_palette <- unname(palette.colors(4, "Okabe-Ito"))

names <- c("month", "year", "full_time", "part_time", "total")
american <- read_table("data/american_airline_empl.txt",
                       skip = 1, col_names = names)
delta <- read_table("data/delta_airline_empl.txt",
                    skip = 1, col_names = names)
federal <- read_table("data/federal_express_empl.txt",
                      skip = 1, col_names = names)
united <- read_table("data/united_airline_empl.txt",
                      skip = 1, col_names = names)

# start with a named list to recover the names when binding rows
employees <- list(american = american, delta = delta,
                   federal = federal, united = united) |>
    bind_rows(.id = "company") |>
    mutate(month = make_date(year = year, month = month)) |>
    select(-year)

full_names <- list(american = "American Airlines",
                   delta = "Delta Airlines",
                   federal = "Federal Express",
                   united = "United Airlines")
job_type <- list(full_time = "Full-time",
                 part_time = "Part-time",
                 total = "Total")

lapply(names(job_type), function(idx) {
    employees |>
    ggplot(aes(x = month, y = .data[[idx]], colour = company)) +
        geom_line(lwd = 0.8) +
        scale_colour_manual(values = my_palette, labels = full_names) +
        labs(x = "Year", y = "Number of employees", colour = "Company",
             title = paste(job_type[[idx]], "Employees"))
})

minmax <- employees |>
    group_by(company) |>
    summarize(min = min(total), when_min = month[which.min(total)],
              max = max(total), when_max = month[which.max(total)])

employees |>
    mutate(part_time = 100 * part_time / total) |>
    ggplot(aes(x = month, y = part_time, colour = company)) +
        geom_line(lwd = 0.8) +
        scale_colour_manual(values = my_palette, labels = full_names) +
        labs(x = "Year", y = "Part-time employees (%)", colour = "Company")

employees |>
    dplyr::filter(year(month) > 2018) |>
    ggplot(aes(x = month, y = total, colour = company)) +
        geom_line(lwd = 0.8) +
        geom_hline(aes(yintercept = min, colour = company,
                       linetype = "Historic minimum"),
                   minmax, lwd = 0.6) +
        scale_linetype_manual(name = "", values = rep("dashed", 4)) +
        scale_colour_manual(values = my_palette, labels = full_names) +
        labs(x = "Year", y = "Number of employees", colour = "Company")

flights$date <- as_date(flights$time_hour, tz = "EST")

flights |>
    count(date, origin) |>
    ggplot(aes(x = date, y = n, colour = origin)) +
        geom_point(size = 0.8) +
        geom_smooth(aes(fill = origin), lwd = 0.8) +
        scale_colour_manual(values = my_palette) +
        scale_fill_manual(values = my_palette, guide = "none") +
        labs(x = "Date", y = "Number of flights", colour = "Airport") +
        guides(colour = guide_legend(override.aes = list(fill = NA)))

flights |>
    mutate(day_type = ifelse(wday(date) %in% c(1, 7),
                             "weekend", "workday")) |>
    # count by day, keeping info on week and day_type
    count(week = week(date), day, day_type) |>
    # regroup by week and day_type to get the averages we want
    group_by(week, day_type) |>
    summarize(avg = mean(n)) |>
    ggplot(aes(x = week, y = avg, colour = day_type)) +
        geom_line(lwd = 0.8) +
        scale_colour_manual(values = my_palette,
                            labels = c("Weekends", "Workdays")) +
        labs(x = "Week", y = "Average flights per day",
             colour = "Day type")

dep_delays <- flights |>
    # some flights don’t have a recorded departure and/or arrival time
    drop_na(dep_delay) |>
    group_by(date, origin) |>
    summarize(avg = mean(dep_delay),
              min = min(dep_delay),
              max = max(dep_delay))

plot_delays <- function(key) {
    names <- c(
        "Maximum" = "max",
        "Minimum" = "min",
        "Average" = "avg"
    )

    ggplot(dep_delays, aes(x = date, y = .data[[names[key]]],
                           colour = origin)) +
        geom_point(size = 0.8) +
        geom_smooth(aes(fill = origin), lwd = 0.8) +
        scale_colour_manual(values = my_palette) +
        scale_fill_manual(values = my_palette, guide = "none") +
        labs(x = "Date", y = paste(key, "departure delay (min)"),
            colour = "Airport") +
        guides(colour = guide_legend(override.aes = list(fill = NA)))
}

plot_delays("Average")

plot_delays("Maximum")

plot_delays("Minimum")

flights |>
    drop_na(air_time) |>
    # distances are in miles, air_time is in minutes
    mutate(speed = distance * 1.60934 * 60 / air_time) |>
    group_by(date) |>
    summarize(avg = mean(speed), sd = sd(speed)) |>
    ggplot(aes(x = date, y = avg)) +
        geom_ribbon(aes(ymin = avg - sd, ymax = avg + sd),
                    fill = "#627bb4", alpha = 0.5,
                    colour = "transparent") +
        geom_line(colour = "#4a417b", lwd = 0.8) +
        labs(x = "Date", y = "Average speed during the day (km/h)")

flights |>
    count(date, carrier) |>
    slice_max(n = 2, by = date, order_by = n) |>
    head(10) |>
    kbl()

flights |>
    count(week = week(date), carrier) |>
    slice_max(n = 2, by = week, order_by = n) |>
    head(10) |>
    kbl()

flights |>
    count(month, carrier) |>
    slice_min(by = month, order_by = n) |>
    mutate(month = month.abb[month]) |>
    kbl()

flights |>
    group_by(month) |>
    summarize(carrier = carrier[which.max(distance)],
              max_dist = max(distance)) |>
    mutate(month = month.abb[month]) |>
    kbl()

flights |>
    dplyr::filter(distance == max(distance)) |>
    distinct(origin, dest) |>
    kbl()